You can download and use any LLM in GGML format, although the parameters for the model may have to be tweaked accordingly. I've primarily used Llama-2-7b (13b and 70b should work just as well).
- Download an LLM in GGML format
- Define the environment variable
MODEL_PATH
(in a .env file) to be the path to the model you want to use - Run
npm install
- Run
npx tailwindcss -i ./input.css -o ./style/output.css --watch
- Run
cargo leptos watch
- Navigate to
localhost:3000
- Start chatting!